www.phidot.org

Just hoping for a quick clarification and recommendation to report my results. I ran known fate survival estimation. I had 3 encounters to give me 2 periods that I was calculating the survival estimate for. Those periods were 3 months in the summer and then 8 months over winter.

My question is regarding how to report monthly survival or the yearly survival. The estimate of apparent survival is 0.98. Which I believe is the estimate to survive the entire observation period (11 months). So, if I were to report the monthly survival is it reasonable to take 0.98^3 and 0.98^8 and talk about the seasonal monthly estimates, or simply take 0.98^11 to get averaged monthly survival?

Thanks for your suggestions!

kernicholson wrote:Just hoping for a quick clarification and recommendation to report my results. I ran known fate survival estimation. I had 3 encounters to give me 2 periods that I was calculating the survival estimate for. Those periods were 3 months in the summer and then 8 months over winter.

So, for a time-dependent model, you have 2 reported estimates -- one for 'summer', and one for 'over-winter'.

My question is regarding how to report monthly survival or the yearly survival. The estimate of apparent survival is 0.98. Which I believe is the estimate to survive the entire observation period (11 months).

.

No. Not sure why you're referring to only a single estimate -- you should have 2 (noted above): 'summer' and 'over-winter'.

More to the point, for any given estimate, you need to specify the interval before you run your analysis. If you do this correctly, for (say) summer, and tell MARK (or whatever software you're using -- I'll assume MARK) that the your intervals are 3 and 8, then MARK will report the

and

respectively -- meaning you'll get a monthly estimate for summer, and a monthly estimate for 'over-winter'.

So, if I were to report the monthly survival is it reasonable to take 0.98^3 and 0.98^8 and talk about the seasonal monthly estimates, or simply take 0.98^11 to get averaged monthly survival?

Thanks for your suggestions!

If you set the interval correctly in MARK, then it is already reporting your 'average' monthly value (not a true 'average' -- merely the n-th root of the interval over n months).

You really need to read the -sidebar- beginning on p. 25 of Chapter 4 of the MARK book:

http://www.phidot.org/software/mark/doc ... /chap4.pdf

Thank you so much for a speedy response.

Not sure why you're referring to only a single estimate -- you should have 2 (noted above): 'summer' and 'over-winter'.

More to the point, for any given estimate, you need to specify the interval before you run your analysis. If you do this correctly, for (say) summer, and tell MARK (or whatever software you're using -- I'll assume MARK) that the your intervals are 3 and 8, then MARK will report the

and

respectively -- meaning you'll get a monthly estimate for summer, and a monthly estimate for 'over-winter'.

You are correct, I did input 3 and 8 in the initial start up screen for the intervals for encounter occasions and therefore I do get 2 Phi's. I wrote my question poorly, I apologize.

If I want to know the survival probability to survive till the end of the survey period I would have to start all over, input 3/11 for summer and 8/11 for winter and then re-run the analysis. I was given the suggestion to (without rerunning everything) take the Phi result and ^12, which is what initially got me confused because I have 2 Phi's - which one would I use? The only way I figure that can work is if I hold season constant, get 1 value of Phi and then ^12.

If I want to know the survival probability to survive till the end of the survey period I would have to start all over, input 3/11 for summer and 8/11 for winter and then re-run the analysis. I was given the suggestion to (without rerunning everything) take the Phi result and ^12, which is what initially got me confused because I have 2 Phi's - which one would I use? The only way I figure that can work is if I hold season constant, get 1 value of Phi and then ^12.

Delta method -- Appendix B. Very first example for multivariate applications -- p. 11.